2014 IEEE International Symposium on Multimedia

Graphical Programming Environments for Educational Robots:

Open Roberta - Yet another One?

Beate Jost, Markus Ketterl, Reinhard Budde, Thorsten Leimbach

Fraunhofer IAIS
Schloss Birlinghoven
53754 Sankt Augustin, Germany
{Beate.Jost, Markus.Ketterl, Reinhard.Budde, Thorsten.Leimbach}
@iais.fraunhofer.de

Abstract system have been invented or enhanced during the last

years. But did they really take into consideration what
In recent years, an increasing number of school teachers, the main multipliers, really need - especially
children is beginning to learn about robotics in the in the classroom?
classroom in order to stir their interest in STEM Nowadays interdisciplinary teaching is the way teach-
professions. Teachers rely on simple educational robots ers promote the school children’s technical competence.
and intuitive programming environments and graphical These teachers are not necessarily math, physics or
programming environments have become a frequent computer science experts. Quite often they are respon-
starting point for young robotics newbies. However, sible for different school subjects or are primary school
currently available tools do often not sufficiently support teachers. Certainly some of the interested teachers al-
teachers and students in the classroom. In this study, ready know how to interact successfully with different
we evaluate programming environments for educational (education) robotic platforms, programming languages
robots; our results point to the need of lowering the com- and corresponding coding environments but there is a
plexity of tools as well as of incorporating combinations significant amount of teachers without basic knowledge.
of web and cloud technologies, embedded systems and With the Roberta - Learning with Robots initia-
communication concepts into these environments. The tive Fraunhofer IAIS has developed and implemented a
technical part of this work presents Open Roberta – an growing concept to assist lectures and teachers directly
open source based addition to commercial educational in the classroom with workshops and practical re-usable
robot environments that addresses these needs. tutorials back in 2002. These hands-on programming
and guiding materials are growing on a daily basis and
can be applied directly in a curriculum. Roberta espe-
1 Introduction cially targets the lack of engineers in general but with a
focus on female engineers in Germany and other Euro-
pean countries by raising children’s interest in technical
The Science, Technology, Engineering and Math professions [5]. With Roberta, annually over 35.000
(STEM) domain is growing rapidly in our working world pupils participate in the program and more then 250
but children leaving school today are rarely interested teachers are becoming certified instructors each year.
in technical professions. There are many different at- These teacher trainings are being evaluated regularly in
tempts to overcome this barrier. One increasingly used order to improve the program and adjust to changing
approach is learning with educational robots in class- demands (see table 1).
rooms. Researchers and developers have done a com-
petent job since the first inception of classroom turtle V. good Good Satisfying Not Satisfying Total
programming back in the 1970’s [16], [7]. Kelleher and 67,94% 29,56% 2,50% 0% 100%
Pausch [10] have already worked out that Educational 462 201 17 0 680
robots are highly motivating for boys and girls hence
keeping this motivation on a high level for children and Table 1. Results of the Roberta teacher train-
for teachers is essential. Manifold educational robotic ing evaluation (from 2008-2013) [11]

978-1-4799-4311-1/14 $31.00 © 2014 IEEE

978-0-7695-5437-2/14 381
DOI 10.1109/ISM.2014.24
 Roberta and the developed programs rely on pro- daily basis? The practical part of this work (section 4)
grammable robots (mainly LEGO hardware) and ad- addresses this identified benefits and system limitations
ditional technology which can be used by pre-school that have been identified and proposes an online based
pupils up to expert college students. The basis is to use graphical programming environment with real roboter
off-the-shelf robot kits and give children and students hardware connection for teachers and learners. This
a target-oriented use of computers without neglecting system can be used in todays web browsers without
the playful element. Within the Roberta network a additional plug-ins or other installations. The presented
special concept has been developed. It starts with Open Roberta platform allows its users to attach their
supervised gender specific robot courses and personal own robot hardware to a computer or tablet device
training workshops for teachers until a certain level and directly download and run the generated online
of expertise is reached. The ideal completion of this programs. Finally section 5 briefly summarizes the im-
effort is the establishment and founding of an regional plementation so far and provides an outlook of intended
Roberta center (RobertaRegionalCenter) which is able future work.
to keep up with further workshops and mentoring (see
[12] for more information). Figure 1 depicts an overview 2 Related work and existing graphical
of founded regional centers in Europe. programming environments for edu-
cational robots

Psychologists and mathematicians Seymour Papert

mentions in his work back in 1981 [16] that children
and young people need to dominate the computer and
not vice versa. Children may not be trained to become
an answering machine. Papert further points out, that
kids must be in a position to easily experiment with
computers and invent new things. The idea of Papert,
scientific thinking and acting through independent re-
search with the wealth to promote each other linking
perception and action in the physical space, continued
and expanded at the MIT in the late 1990s mainly due
to work by Mitchel Resnick, Robbie Berg and Michael
Eisenberg [19].
Powers at al. [17] provided a broad overview about
robot programming systems in 2006. They divided
these systems into automatic programming and manual
programming environments which can be text-based
and/or graphically driven. This basic classification
for graphical programming includes also tools for in-
Figure 1. Roberta regional centers in Ger- teracting with educational robots. The study lacks a
many and Europe deeper insight into possible system couplings and setup
information regarding the connection of real robots,
The remaining course of the paper is organized as fol- simulations or program deployments.
lows: The next chapter presents related work and a brief Other work targets the programming task itself as
characterization of benefits of graphical programming being presented in work from Kelleher and Pausch [10]
with a special focus on classroom teaching. System com- in 2005. The authors provide a detailed taxonomy
monalities, merits and limitations are being addressed and note, that graphical programming environments
in section 3. A focal point is the evaluation of graphical belong to the group of so-called alternatives to typing
programming environments. Particularly the process programs without further information. Major, Kyriacou
of required preparation work and final execution work- and Brereton’s report [13] from 2012 provides a pure
flows which enable user to run their programs on actual literature review of teaching novice programming using
robot hardware is of interest in this study. Furthermore, robots. Although this is a very detailed survey with
are available graphical programming environments and an almost entire list of programming tools the authors
available workflows capable of satisfying the needs of did not make a distinction between text-based and
secondary school level children and their teachers on a graphical programming environments. Moreover pure

382
scientific and/or statistical information seems to be less ating systems? Who can do the installation for one or
important for practical users like teachers, parents or more classrooms? What do we have to pay? How long
other instructors in the educational domain. will this system be supported? How do we select the
For most of the children and some teachers in the right target robot hardware (e.g. costs, stability)? How
classrooms robot programming is the very first contact do we interact with the robots, taking administration
with computer languages and real programming. As restrictions at our school into account?
learning a textual programming language can be really In summary one can say that it is important to lower
stressful graphical pendants are frequently chosen, espe- the installation complexity in general and to provide
cially in the context of educational robotics.Graphical generic open tools that can be used across devices and
programming languages provide advantages for begin- allow an interaction with different robot platforms. Cur-
ners compared to text based code interpretations. One rently this demand can’t be fully satisfied by available
can summarize positive attributes such as a better tools, frameworks and development kits as we highlight
user experience [4], an easier entrance to programming in the next subsection.
paradigms [9, 17] and they are helpful to prevent typi-
cal syntactical errors due to the fact that the syntax of 3.1 Commonly used graphical robot pro-
the statements is already implemented into the visual gramming environments
shapes of the statements [10, 21].
O’Hara and Kay [15] (2003) have not evaluated
Based on our experience we’ve selected some of the
graphical programming environments but educational
most frequently used educational robot ecosystems for
robotics. They highlight the importance of easily avail-
an elaboration; taking into account the previously men-
able tools as a very important element for realizing
tioned criteria. Particularly we are taking a closer look
undergraduate classroom projects. And still, ten years
at the systems:
later, the setup, hardware communication and instal-
Enchanting [1], EV31 [22], GRAPE [6], miniBloq [2],
lation roundtrip is not classroom ready as frequently
Modkit [14], ROBO pro [8], NXT-G1 [23], RobotC2
criticized by Roberta network participants. Some of
[20], RoboMind [18], Ardublock [3].
these limitations are being further addressed in the next
Table 2 depicts different features of the systems under
section, followed by a brief systems in use overview.
evaluation. All of them provide integrated development
environments (IDEs) for multiple computer platforms
3 Merits and common limitations of and operation systems and also present graphical pro-
available systems gramming alternatives in addition to a pure code based
representation to users. Certain robot hardware can be
Teachers and instructors are oftentimes left alone connected and user generated programs can be down-
with technical problems in the classrooms. They loaded from the IDE to the robot device. Some of the
are quite overwhelmed by technical preparation tasks IDEs provide a web interface like Modkit and RoboMind
like system installations for manifold platforms, net- but extra system specific software needs to be installed
work/device setups, firewall restrictions and/or huge for a robot communication. All, except two products
data downloads before they even can start teaching. (Enchanting and Minibloq) are fee-based, users need
Technical resources are limited or oftentimes completely to pay for at least the educational version. In general
missing at schools. Moreover there is a mixture of each of the IDEs is designed for a certain robot system
old/new computer systems, operation systems and avail- or hardware family. ROBOTC, Modkit, and Minibloq
able tools and devices. Many schools have started to support at least two robot platforms. EV3 supports its
focus on using tablet devices to reduce the administra- preceding robot system NXT partially at the moment.
tive overhead. But this technical changeover does not The ideal system candidate needs to include the follow-
play well with the current status of available educational ing features: Visual graphic programming support, fully
robotic environments. web based with possibilities to run the infrastructure re-
Several different environments are being offered in motely or on local school servers, connections to robots
the context today. In most cases they belong to or should be easy (e.g. WLAN, Bluetooth) and it should
are designed for a specific robotic system. To choose not be restricted to a certain vendor, the technical
the right one out of them for a specific school or learn- complexity like compilation and program preparation
ing task is not easy. A lot of questions have to be should run in the background, price sensitive (or free) in
answered before a decision for a system can be made. 1 LEGO
R MINDSTORMS
R Software
Like for example: What computer systems do we use 2 forLEGOR MINDSTORMSR and MINDSTORMS with
and maintain? Is the software available for our oper- TETRIXTM

383
 Platforms limitations. In order to solve this recurring issues we

Mac OS X
have started an open source network initiative with the

Windows
IDE Costs Robot intention to make the programming of robots easier

Linux
and more fun for all participants. A major goal of the

web
Open Roberta project is to work on a software platform
dissemination for a scalable, education-friendly, cloud-
NXT-G × - × - (yes)1 12 based robot development kit.
EV3 × - × - (yes)1 2
Enchanting × ×3 × - no4 1 4.1 From desktop to tablet towards an
RobotC × - - - yes 2 easy robot program deployment
ROBO pro × - - - yes 1
Modkit × - × (×)5 yes 2 We have a deep interest in software portability.
GRAPE × - - - yes 1 Portability stands for reusability in schools e.g. robots
miniBloq × ×6 - - no4 2 have been acquired in one year and computers for the
RoboMind × × × (×)5 yes 1 classroom in another year. Can the robots still be
programmed with the newly purchased tablet pc’s?
Ardublock × × × - no4 1
And what about the other way round? Cross-platform
Table 2. IDE comparison of portability, costs IDE’s are well suited in this case, but none of the given
and number of supported robots programming environments are really cross-platformed.
Modkit and RoboMind have a web interface but the real
1 only
work (compiling the program) still has to be done on
the educational version is fee based
2 third-party the computer and therefore special drivers and software
sensors and motors are supported
3 experimental version is necessary. “Cloud programming” or programming via
4 open source a web interface on a server would make administration
5 needs further platform depending software for the
redundant: no installation effort, updates are installed
robot ↔ application communication
6 preliminary version automatically, available also for tablets.

4.2 Technical perspective and goals

order to allow schools to easily join, test and participate
as individual or as a group (school class). None of the The age of the students we are mainly trying to
available products currently fulfill this criteria without address is between 10 and 16. The teacher may have
technology getting in the way. a computer science background but it is much more
This is why we are proposing a fully online, web likely that his or her main experience is natural science
based alternative for visual educational robot program- or mathematics. The preparation time for a course
ming which allows to easily use the browser for pro- is limited, the hardware heterogenous, professional IT
gramming without manual installation and preparation support is almost never available and the number of stu-
with an easy connection with different robot hardware. dents in a course is usually large. This determines our
The compilation process and program preparation is goals for a programming environment for educational
intended to run on local or cloud based servers and is robots.
not limited to a certain computer platform or system The intended project will be fully based on open
software. source technology. The development process shall be
open and flexible which includes involving extern parties
4 Open Roberta - An open source on- like interested partners within the Roberta network (e.g.
also universities) or other interested individuals. An en-
line programming environment for vironment must have almost no setup, no configuration,
teaching and learning with flexible in- it has to be available for almost all platforms/operating
tegration of robot hardware systems and must be easy to use, because only very
well known user interface concepts are used. We believe,
Parts of the Roberta workshop material is currently that in this case using a browser is the best decision to
based upon existing proprietary LEGO Mindstorms IDE abstract from the underlying operating system.
software solutions which are available only for Windows- The browser’s user interface is well known to almost
and Mac computers. Roberta teachers are frequently all teachers and students, including its limitations. Fur-
faced with the previously mentioned roadblocks and thermore it is probably the only solutions that allows

384
the use of the diverse tablets. Thus a client server design development would hide the comprehensive possibili-
is appropriate which is based on: The design of a server ties of computer science. Furthermore, if problems (and
back end, that relies on Ajax and REST. We favor Java solutions) become bigger and bigger, graphical program-
and its many robust well known frameworks for this job. ming evolves to a tough job very often. Thus we plan
The use of a Javascript enabled browser and adopt some to offer more perspectives of programming especially a
well engineered frameworks for the fronted architecture. round trip between graphical and textual representa-
Of course we cannot avoid a basic robot setup, because tion for the user. The programming environment must
we need a communication facility between server and support different code generators, too, Java being the
robot, but – compared to work station setups – this first. The generated code has to be human readable,
should be easy and straightforward. thus the many levels of abstraction in system develop-
After having decided to use a browser-based client, ment can be experienced by the pupils. Last, but not
the communication between robot and client is the least, execution of program on a robot as well as on a
major problem, as the client is executed in a ”sandbox”. simulation engine should be possible.
Any local solution, be it Bluetooth, NFC, direct USB It is almost impossible to foresee the way educational
connection is not feasible without heavy configuration software will be used in the future (this is true for other
and administrator privileges.The only practical solution kinds of software as well). It is very likely, that people
to this problem is, that both the browser client (using benefit from the evolution of the software into many
Ajax calls) and the robot (acting as a HTTP client) are very distinct directions. We don’t expect to have the
connected to the same server using Wifi. In most cases resources to achieve that on our own. Therefore our
this will be based on a Wifi network that is connected programming environment has to be open source. Fur-
to the internet. Then any public server hosting the thermore it should not confuse developers enhancing
backend software can arrange the connection between it. We want to achieve that by applying well-known
robot and browser client. standards (rules as design by contract, tools as sonar)
A simple and sufficiently safe solution would be a and well-known frameworks (as jetty, jersey, jaxb2, hi-
token, that can be used to pair the robot and a browser bernate, antlr4). The code base should be as small as
client. The token could be generated on the robot and possible. The NIH (not invented here) syndrome must
retyped by the user into the browser window. This be avoided. Using existing frameworks is preferred to
establishes the connection. Such a connection exposes re-inventing solutions. We hope to have tackled most
user names, passwords, programs and information about of our goals in the first version of our programming
teachers and pupils to the public. In the straight past environment.
this would have been advertised as a kind of cloud
server and cloud storage. But many schools, teachers 5 Conclusion and outlook
and students have concerns w.r.t. to privacy. They
should not be forced to expose their data when using The presented Open Roberta project will have a huge
the programming environment. Therefore it should impact for the existing Roberta network. The inten-
be possible to use any local Wifi network to which a tion is to solve many of the frequently occurring issues
computer and a robot can connect to. A local server reported by our certified educational robotic teachers
must be deployable on any -even small- notebook or (Roberta teacher) in the actual classrooms. We’ve ex-
PC in this network. A minimal solution consisting out emplified merits and common limitations of available
of a smartphone establishing a hot spot, a notebook on visual programming development environments with
which the server is deployed and (the same or more) a special focus on technical obstacles. This overview
notebooks or tablets executing the browser client should and analysis is the baseline for the development of the
be usable without any restriction. open source visual online programming environments
Actually we expect that a Java >= 7 installation for educational robots as presented in this work. The
on a computer is the only prerequisite for deploying further development of Open Roberta is going to fo-
the back end part of our programming environment. cus on three main traces. A further extension of the
The client assumes no installation - beside a Javascript visual programming language including adaptation of
enabled browser - at all. To attract rookies it makes additional robot sets. A collaborative programming
sense to offer a pure graphical programming language. environment for complete school classes and/or individ-
We have chosen to use blockly for that purpose, as uals. The extension of available tutorials and hands-on
many other projects did. But it would be a bad idea to practical guides for both teachers and kids with an in-
limit ourselves to that kind of program representation. tegration of blended learning concepts, tools and online
Offering only one perspective to understand software infrastructure.

385
Acknowledgements [15] K. J. O’Hara and J. S. Kay. Investigating open source
software and educational robotics. Journal of Comput-
ing Sciences in Colleges, 18:8–16, 2003.
Besides the members of the Open Roberta commu- [16] S. Papert. Mindstorms: Children, computers, and
nity we would like to thank Google for their generous powerful ideas. Basic Books, New York, NY, January
project support and advice. 1981.
[17] K. Powers, S. Cooper, K. J. Goldman, M. Carlisle,
M. Mcnally, and P. Viera. Tools for Teaching Intro-
References ductory Programming : What Works ? ACM SIGCSE
Bulletin, 38:560–561, 2006.
[1] Enchanting. Available online: http://enchanting. [18] Research Kitchen. RoboMind, 2014. Available online:
robotclub.ab.ca/tiki-index.php, accessed 08-22-2014. http://www.robomind.net/en/index.html, accessed 08-
[2] Minibloq, 2014. Available online: http://blog.minibloq. 26-2014.
org/, accessed 08-25-2014. [19] M. Resnick. Turtles, termites, and traffic jams - explo-
[3] Ardublock. Ardublock a Graphical Programming Lan- rations in massively parallel microworlds. MIT Press,
guage for Arduino, 2014. Available online: http: 1998.
//blog.ardublock.com/, accessed 08-15-2014. [20] Robomatter, Inc. ROBOTC for LEGO MIND-
[4] T. Booth and S. Stumpf. End-User Experiences of STORMS, 2014. Available online: http://www.robotc.
Visual and Textual Programming Environments for net/download/lego/, accessed 08-25-2014.
Arduino. In Y. Dittrich, M. Burnett, A. Mø rch, and [21] E. Rosenbaum, E. Eastmond, and D. Mellis. Empower-
D. Redmiles, editors, End-User Development, pages ing programmability for tangibles. In TEI ’10: Proceed-
25–39. Springer Berlin Heidelberg, 2013. ings of the fifth international conference on Tangible,
[5] A. Bredenfeld and T. Leimbach. The Roberta Initiative. embedded, and embodied interaction, pages 357–360,
Proceedings of the 2nd International Conference on New York, New York, USA, 2010. ACM Press.
SIMULATION, MODELING and PROGRAMMING [22] The LEGO Group. LEGO R MINDSTORMS R

for AUTONOMOUS ROBOTS (SIMPAR 2010), pages Education EV3-Software, 2014. Available online:
1–2, Darmstadt, 2010. http://education.lego.com/de-de/lego-education-
[6] S. Enderle. The qfix Robot Kits. In D. Gottscheber, product-database/mindstorms-ev3/2000045-lego-
Achim and Enderle, Stefan and Obdrzalek, editor, Re- mindstorms-education-ev3-software-single-user,
search and Education in Robotics — EUROBOT 2008, accessed 08-25-2014.
pages 84–95. Springer Berlin Heidelberg, 2009. [23] The LEGO Group. LEGO R MINDSTORMS R
Education NXT Software 2.1.6 (inkl. Mess-
[7] W. Feurzeig. Educational Technology at BBN. Annals
werterfassung), 2014. Available online:
of the History of Computing, IEEE, 28:18–31, 2006.
http://education.lego.com/de-de/lego-education-
[8] Fischertechnik GmbH. ROBO pro, 2014. Available
product-database/mindstorms/2000080-lego-
online: http://www.fischertechnik.de/, accessed 08-15-
mindstorms-education-nxt-software-v-2-0, accessed
2014.
08-25-2014.
[9] Z. Istenes and A. Pásztor. Using Innovative Devices in
the Education of IT from Primary to University Level.
In Sixteenth International Electrotechnical and Com-
puter Science Conference, ERK, volume 24, page 26,
2007.
[10] C. Kelleher and R. Pausch. Lowering the Barriers to
Programming : A Taxonomy of Programming Environ-
ments and Languages for Novice Programmers. ACM
Computing Surveys (CSUR), 37(2):83–137, 2005.
[11] T. Leimbach, B. Jost, U. Petersen, J. Börding, and
S. Härtig. Roberta-Grundlagenband EV3. Fraunhofer
Verlag, Stuttgart, Germany, 2014.
[12] T. Leimbach and U. Petersen. Roberta-Goes-EU: Ziel-
gruppensensible Robotikkurse als Erfolgsmodell in Eu-
ropa, pages 183–194. Gender und Diversity in den
Ingenieurswissenschaften und der Informatik. Univer-
sitätsVerlagWebler, Bielefeld, 2008.
[13] L. Major, T. Kyriacou, and O. Brereton. Systematic
literature review: teaching novices programming using
robots. IET Software, 6(6):502–513, 2012.
[14] Modkit LLC. Modkit for VEX, 2014. Available online:
http://www.modkit.com/vex, accessed 08-25-2014.

386